The Thermodynamic Ionization Constants of Carbonic Acid

The Thermodynamic Ionization Constants of Carbonic Acid. By Duncan A. MacInnes and. Donald Belcher. Due to their importance in the fields of biology, ...
1 downloads 0 Views 1MB Size
2630

DUNCAN A.

[CONTRIBUTION FROM

THE

MACINNES AND

DONALD BELCHER

Vol. 55

LABORATORIES OF THEROCKEFELLER INSTITUTE FOR MEDICAL RESEARCH]

The Thermodynamic Ionization Constants of Carbonic Acid BY DUNCAN A. MACINNES AND DONALD BELCHER Due to their importance in the fields of biology, geology and inorganic chemistry the ionization relations of carbonic acid have received a large amount of investigation. The results of different researches, however, while agreeing among themselves in the orders of magnitude of the two ionization constants, show considerable diversity in the actual figures. The variation is especially great when the published values obtained by potentiometric and conductometric methods are compared. Since the most recent determinations of these important constants two new methods of research have become available. These are galvanic cells without liquid junctions, as developed by Harned and associates? and glass electrodes which can be used in precision measurements. Using these methods we have redetermined the first and second dissociation constants of carbonic acid, attaining, we think, higher accuracy than has been reached in earlier researches. The following pages describe determinations a t 25'. Further measurements a t 38' will be carried out.

The First Dissociation Constant of Carbonic Acid The primary dissociation of carbonic acid follows the reaction COz

+ H20 L_ HzCOs

H+

+ HCOs-

and has the dissociation constant

in which the parentheses represent activities, and the brackets concentrations, of the constituents enclosed. The terms referring to carbon dioxide are here, and in what follows, considered to refer to that substance in solution both as COz and HzC03.' The y values are activity coefficients of the constituents represented by the subscripts. The method used in this research to obtain the constant K , has been to measure potentials of cells of the type Ag, AgC1, KHC03, KC1, COS(dissolved), glass, buffer f KC1, AgCI, Ag

(A)

in which the carbon dioxide in the solution is in equilibrium with the gas a t a known partial pressure. The reason for the choice of this particular arrangement will be evident after a consideration of the theory of the operation of the cell. (1) The constant KI. which is strictly equal to (H+)(HCO8-) /(COz 4-HzC03). has been shown by Faurholt [ J . chim. phys., 21, 400 (1924)] to be equal to KI = K t i ( l / K h 4- 1) in which Kt (H (HCOa-) /(HaCOa), the "true" ionization constant of &Cos, and K h = (HzCOa) /(Cor). Since kh is a constant if the activity of the solvent water does not change, KI will also be a constant with this same limitation. Thiel and Strohecker [Ber., 47, 945 (1914)l and Pusch [Z.Elekrrochem., 22, 206 (1916)]have shown that less than 1%of the dissolved carbon dioxide is in the hydrated form HCOa.

.=

j)

July, 1933

THETHERMODYNAMIC IONIZATION CONSTANTS OF CARBONIC ACID 2631

Except for a constant, the portion of the cell glass, buffer

+ KCl, AgCI, Ag

can be considered to be a hydrogen electrode, i. e., one that yields a potential depending upon the hydrogen-ion activity of the solution in which it is immersed. Cell A can thus Be treated as if i t were of the type Ag, AgC1, KIfCOs, KC1, COz, Hz

(C)

Ag, AgC1, C1-, H+, Hz

(D)

or more simply with, however, the practical advantage that hydrogen gas is not necessary for the measurements. The potential of cell D is given by the expression E == Eo

+ ( R T / F )In (H+)(Cl-) = E O+ (RT/F)(ln [H+][Cl-] + In YHYCI) (2)

Substituting [ H + ]from equation 1 we have

As very good approximations we may assume (a) ycO2to be unity and (b) the concentration of carbon dioxide [COz] to follow Henry’s law [Con] = SPCO,

(3a)

in which Pco, is the partial pressure of carbon dioxide and S is a constant. The validity of these assumptions will be discussed later in this article. Equation 3 can thus be put in the form E = E o + ( R T / F ) In [Cl-l/[HC03-1 -I- ( R T / F ) lp Y H Y C I I Y H Y H C O ~4( R T / F )In KlSPco2 (4)

By choosing [Cl-] = IHC03-1 and recalling that the use of the glass electrode in cell A involves another constant, which we will call E,, equation 4 can be recast in the form in which E A is the potential of Cell A and pK1 is the negative logarithm of the thermodynamic ionization constant. The value of pK: approaches pK, as the products YHYCl and Y H Y I I C ~approach ~ unity a t infinite dilution. The experimental portion of this part of the research consisted in the measurement of the potentials of cells of type A a t a series of ionic strengths. From these data a value of the ionization constant K1 was obtained by an extrapolation which will be discussed below. A study was also made of the effect of varying the partial pressure of carbon dioxide on the potential of the cell. The Apparatus Used in Measuring the First Ionization Constant The form of cell adopted after considerable experiment is shown in Fig. 1. The vessel A has four openings for closely fitting rubber stoppers through which pass the electrode tubes. The tube marked E carried a silver-silver chloride electrode. It was open a t the lower end, and contained the same solution as that in vessel A. The tubes G, Gr’ and G “ are “glass electrodes,” that is t o say, they are half cells with glass membranes a t the points indicated by m in the diagram. The construction and use of these

2632

DUNCAN A . MACINNES A N D DONALD BELCHER

Vol. 55

electrodes has been fully described in a series of papers from this Laboratory.* Such electrodes function, with careful use, reversibly to the hydrogen-ion activity, to *0.1 mv. within the PH range 1 to 8. The electrodes used in this research differed from those previously described in two details. For one thing the solution contained in the mounting tubes and into which the silversilver chloride electrodes b, b' and b" were inserted consisted of an acetate buffer solution containing potassium chloride (approximately 0.01 N acetic acid, 0.01 N potassium acetate and 0.01 N potassium chloride), instead of the 0.1 N hydrochloric acid used in our earlier work. The other change was the enlargement shown in the upper end of the mounting tube, which was made of Jena thermometer glass 16 111. This made it possible to mount the silver-silver chloride electrodes on rubber stoppers. The sensitive surface of these electrodes was thus prevented from coming into contact with the wall of the tube. This practice resulted in a marked improvement in the steadiness of the potential readings. The stoppers for Fig. 1. the electrodes G, G' and G#had capillary openings (not shown) to prevent the building up of pressures which would burst the glass membranes. Preliminary experiments showed unmistakably that cells of type A would not yield definite potentials if the carbon dioxide were bubbled through the solution in the cell or through the liquid in a vessel used to saturate the gas with water vapor. Strictly, equilibrium cannot be reached under these conditions as the gas is not a t a definite pressure. The effect is undoubtedly related to the observations of W i ~ ~ k l eCady, r , ~ Elsey and Berger,' and other workers who have found that too high gas solubilities result from violent bubbling or shaking of a gas with its solvent. To avoid such effects the gas was passed slowly in the direction indicated in Fig. 1, over the surface of the solution in the vessel A after passing over solution of the same composition in the saturator S. The whole apparatus was given a gentle rocking motion around the axis a-a'. After making this change in the design of the apparatus the potentials reached definite reproducible values. As already stated, the use of the glass electrode involves a n additive constant Eg in the computations. Its value for the present case is that of the potential of the cell H2, 0 . 0 1 N HAC, 0 . 0 1 N KAc, 0.01 N KCl, AgCl, Ag (E) the solution in this cell being that in the glass half cell. This potential was determined with great care for the two solutions used for this purpose in this research. These solutions had only approximately the composition indicated. Definite known compositions are not essential as the electrodes are used for reference only. As pointed out in previous papers the potential of the cell E is equal t o that of a cell (2) Macinnes and Dole, Ind. Eng. Chcm., Anal. Ed.,1, 67 (1929); J . Cen. Physid., 14, SO5 (1928-1929); THIS JOURNAL, 62, 29 (1930); MacInnes and Belcher, i b i d . , 69, 3315 (1931). (3) Winkler, Bcr., P4, 89 (1891). (4) Cady, Elsey and Berger, THIS JOURNAL, 44, 1456 (lQZ2).

July, 1933

THE THERMODYNAMIC IONIZATION CONSTANTS OF CARBONIC ACID

2633

Hz, “solution X,” glass, 0 . 1 N HAC, 0 . 1 AVKAc, U. 1 KCI, AgCl, Ag

if the two surfaces of the glass are reversible to the hydrogen-ion activity. This is the case for the glass developed in this Laboratory if “solution X” has a PHvalue between about I and 8. In obtaining the potential, Bg,of cell E the best available silver-silver chloride electrodes were used, and hydrogen of the highest possible purity, and the usual corrections were made for variation of barometric pressure, and for the vapor pressure of the solution. In addition a correction for the “asymmetry potential” existing in the glass membrane must be made. This was found by measuring the e. m. f . of the cell Ag, AgCI, solution G ( o ) ,glass, solution G (i), AgC1, Ag (F) in which “solution G” has the composition of that in cell E and fills the glass half cell. The designations (0)and (i) represent inside and outside that half cell. The silversilver chloride electrodes inserted into “solution G (0)”were selected with special care, as they are the actual reference electrode::. Any variation in the corresponding electrodes in the glass half cell appears as part of the “asymmetry potentials.” The silver-silver chloride electrodes were made with the kind assistance of Dr. Alfred S.Brown, who has been successful in improving their reproducibility. As the details af his study will be published shortly it will be sufficient to state that the electrodes consist of platinum wires, plated with silver from a bath containing no free cyanide, and finally given a coat of chloride by electrolysis. Preparation of Solutions and Experimental Procedure Sodium and potassium chlorides were purified by crystallization from hydrochloric acid solution, followed by centrifuging and drying a t 120”. Before use in preparing the solutions they were fused in vacuum, using the Richards bottling apparatus. Sodium bicarbonate solution was mad(: by passing a slow stream of carbon dioxide through Standardized sodium hydroxide solution until a constant weight was reached. The sodium hydroxide solution was standardized to 0.01yo against pure benzoic acid from the Bureau of Standards, by differential electrometric t i t r a t i ~ n . ~ Potassium bicarbonate was prepared in the following manner. A saturated solution of a good commercial grade of the salt, of German origin, was filtered through sintered glass. The solution was then saturated a t room temperature with carbon dioxide, and a volume of redistilled alcohol equal 1.0 that of the solution added. The mother liquor was then drawn away from the precipitated salt through a sintered glass filter. .4further separation of mother liquor was made with a centrifuge. The salt was finally dried over sulfuric acid in a desiccator through which a stream of carbon dioxide gas was passed. For work to be described later in this paper several weighed samples of the bicarbonate were converted to carbonate by fusion in a n atmosphere of carbon dioxide. The difference between the observed and expected weight of carbonate was never over 0.356. For each stock solution the respective chlorides and bicarbonates were taken in equimolar proportions. For the more dilute solutions used in the cells weighed portions of the stock solution were diluted in standardized volumetric flasks a t 25‘ using conductivity water. The procedure for a determination was as follows. The clean dry cell, A and S of Fig. 1, was thoroughly rinsed with the solution under investigation. The glass electrodes, G, G’ and G”, which had been rinsed with the solution, were then inserted, and a solid rubber stopper was used instead of the stopper carrying the electrode in tube E. The current of carbon dioxide gas was then started and the cell was gently rocked for three hours. The tube E was then filled with the saturated solution in vessel A by ( 5 ) Maclnnes and Cowperthwaite, THISJOURNAL, 63, 555 (1931).

2634

Vol. 55

DUNCAN A. MACINNES AND DONALD BELCHER

closing the outlet of the gas. The insertion of a rubber stopper carrying a silver-silver chloride electrode completed the cell, which, except for the slight solubility of silver chloride, held a solution of uniform composition throughout. After a further rocking of half an hour poteatials were measured a t intervals until the readings were found to be constant. A maximum of four hours from the time of assembly of the apparatus was necessary. The readings were made using the potentiometer and electrometer equipment described by MacInnes and Belcher.6 The apparatus was kept in a constant temperature room the air temperature of which was regulated within h O . 1 of 25'. However, a thermometer placed in a small vessel of water or oil indicated temperature variations of *0.03°. I t is of interest, in connection with the use of the glass electrode, to note that most of the measurements recorded in Tables I and I1 were made in the summer time when the humidity was high. On account of careful insulation, readings could be made with accuracy with relative humidities up to 75%. At still higher humidities some difficulty was encountered.

The Data for Computing the First Ionization Constant In Tables I and 11, which are largely self-explanatory, are given the experimental data used in obtaining the first ionization constant of carbonic acid. The first of these tables gives the results obtained with cells containing potassium bicarbonate and chloride, and Table I1 results with the corresponding sodium salts. In both cases the carbon dioxide gas was a t approximately atmospheric pressure. The second column of the tables gives Ea, the measured potential of the cells of type A, and the fourth gives E A , which is the same potential corrected for the asymmetry potential of the glass electrode. The separate figures for Ea a t each ionic strength are readings of the two or three glass electrodes used. It will be noted that after correcting for the asymmetry potential they yield the same value of E A within 0.1 mv. TABLE I DATAFOR THE COMPUTATION O F THE FIRSTDISSOCIATION CONSTANT O F CARBONIC ACID Solutions: Equal Concentrations of KHCOs and KC1. COS 99.54%. E , = -0.6326 volt. Ionic strength. p

0.002181 .003620

.004281 ,005058

.006814

Ea

-0.0533 - ,0523 - ,0539 - ,0530 - ,0511 - .0539 - .0543 - .0517 - ,0527 - ,0557 - ,0539 - .0537 - ,0546

Asymmetry potential

+o. 0002 - ,0006 ,0009 .0000 - ,0019 .0006 ,0009 - ,0013 - .0005 ,0026 .0007 ,0006 ,0013

+

+ +

+ + + +

EA

-0.0531 - ,0529 - .0530 - ,0530 - .OS30 - ,0533 - ,0534 - .os0 - ,0532 .OS31 - ,0532 - .0531 - ,0533

-

(6) MacInnes and Belcher, THIS JOURNAL, 69, 3315 (1931)

Barometric pressure

observed

computed

762.5

6.339

6,343

761.3

6.342

6.343

756.0

6.344

6.342

761.5

6.345

6.342

754.0

6.342

6.342

OK;

Pk';

July, 1933

THETHERMODYNAMIC IONIZATION CONSTANTS OF CARBONIC ACID 2635 TABLEI

Ionic strength, p

0 01109

,02130

,02139

.02144

Ek

-0.0528 - ,0523 - ,0543 - ,0527 - ,0524 - ,0543 - ,0530 - ,0544 - ,0524 - ,0526 - .0521

,02150

-

,04020

-

,06194

-

,1006

-

,2486

-

,2486

-

,0508 .0539 .0538 ,0500 ,0502 ,0536 ,0522 .0511 ,0504 .0500 .0530 ,0512 ,0527 ,0514 .0503 ,0509

(Concluded)

Asymmetry potential

-0.0002 - ,0008 ,0014 - ,0002 - ,0006 ,0015 0000 ,0014 - ,0006 - ,0004 - ,0008

+

+ + -

,0021

+ .0009 + .0008

- ,0029 - ,0027 ,0007 - ,0002 - ,0013 - ,0019 - ,0024 ,0008 .0000 ,0015 ,0002 - ,0008 - .0002

+ + + +

EA

Barometric PK: pressure observed

PK;

computed

-0.0530

- .0531 - ,0529 - ,0529 - ,0530 - ,0528 - .0530 - .0530 - .0530 - .0530 - .0529 - ,0529 .0530 .0530 - ,0529 - ,0529 - .0529 - .0524 - ,0524 - .0523 - ,0524 - ,0522 - ,0512 - ,0512 - .om2 - .0511 - .0511

-

751 9

6.338

6.342

754.4

6.338

6.340

756.2

6.341

6.340

758.1

6.340

6.340

756.2

6.341

6.340

758.0

6.340

6.338

766.2

6.335

6.336

760.4

6.331

6.331

754.3

6.309

6.314

757.3

6.309

6.314

~

TABLE I1 COMPUTATION OF THE FIRSTDISSOCIATION CONSTANT OF CARBONIC ACID Solutions . .Equal Concentrations of NaHC03 and NaC1. COZ 99.54%. E, = -0.6219 volt.

DATA FOR THE

Ionic strength, ji

0.002942 .004357

,005947

,009967 . 0 1971 04048

Ek

-0.0684 .0685 - ,0631 - ,0629 - ,0629 - ,0684 - ,0679 - .0681 - ,0632 - ,0631

-

-

,0678

,0680 - .0623 - ,0621

Asymmetry potential

+O. 0051

+ ,0051 - ,0003 - ,0004 - ,0005 ,0050 ,0045 ,0047 - ,0002 - ,0002 ,0046 ,0049 - ,0009

+ + +

+ + +

,0010

EA

-0.0633 ,0633 - ,0634 - ,0633 - ,0634 - ,0634 - ,0634 - ,0634 - ,0634 - .0633 - ,0632 - ,0631 - ,0632 - ,0631

-

PK:

PK;

Barometric pressure

observed

765.5

6.337

6.343

766.6

6.341

6.343

772.4

6.345

6.342

769.7

6.344

6.342

764.3

6.337

6.341

763.7

6.335

6.338

computed

2636

DUNCAN A. MACINNES AND DONALD BELCHER

Vol. 55

The Computations Except for a small correction to the results for the more dilute solutions, values of pKi were computed using equation 5. There are two constants that appear in this equation that have been taken from the results of other workers. These are the standard potential of the silver-silver chloride electrode, Eo, and Henry’s law constant, S, of carbon dioxide. A large number of studies have been made for determining the value of Eu. Hitchcock7 has summarized the data up to 1928, and has suggested a convenient method for obtaining this constant from the data. More recent studies are those of Roberts,8 Carmodyg and Harned and Ehlers.Io Hitchcock’s method applies in the concentration range in which the activity coefficient y can be expressed with sufficient accuracy hy the approximate Debye-Hiickel relation

- log y

= a

z/c - BC

(6)

In this expression C is the concentration of a uni-univalent electrolyte and cy ( = 0.5048 a t 25’) and B are constants. The method consists in plotting values of E - 2 RT/F (log C a against the concentration C, E being the potential of the cell Ag, AgCl, HCI (C), Hz, the slope of the resulting straight line being - (2 R T / F ) B and the ordinate a t zero concentration Eo. With the aid of Dr. Alfred S. Brown we have applied this method to the data of the three recent researches mentioned above, using, however, the method of least squares instead of a plot. The results of these computations are to 0.1 mv.: from Roberts’ data Eo = -0.2223, from Carmody’s data Eo = -0.2222, and from Harned and Ehlers’ Eo = -0.2224. We will adopt provisionally the mean value, -0.2223, which agrees with Hitchcock’s value obtained from the earlier data. A redetermination of this constant is in progress in this Laboratory. For Henry’s law constant, S, a t 25’ we have accepted 0.03353 mole/liter atm. which is the mean of the closely agreeing values 0.03357 and 0.03353 from the work of Findlay and associates,11and the value 0.03350 from a paper by KunerthI2interpolated from his results a t 24 and 26’. The partial pressure of carbon dioxide (PcOlin equation 8) has been taken as equal to the barometric pressure corrected for the vapor pressure of water, and for the composition of the gas. The assumptions involved are discussed in the next section of this paper. The gas used in the experiments outlined in Tables I and I1 came from a tank which delivered 99.54 mole per cent. carbon dioxide. The barometric pressures were read on a

+ dc)

(7) (8) (9) (10) (11) (1912); (l?)

Hitchcock, THIS J O U R N A L , 60, 2076 (1928). Roberts, ibid.,63,3877 (1930). Carrnody, ibid., 64, 188 (1932). Harned and Ehlers, i b i d . , 64, 1350 (1932). Findlay and Creighton, .I. Chem. Soc., 97, 536 (1910); Findlay and Shen. ibid., 101, 14.59 Findlay and Williams, i b i d . , 103, 636 (1918); Findlay and Howell i b i d . , 107, 282 (1915). Kiinerth. Phys. Rev.. 19, 511 (1922).

July, 1%:3

THETHERMODYNAMIC IONIZATION CONXANTS

OF

CARBONIC ACID

2fi87

barometer which has been frequently compared with U. S. Weather Bureau standards, and all the usual corrections have been made. Values of PK; computed from our data are given in the next to the last column of Tables I and 11. These values have been computed from equation 5 and their meaning is evident from that equation. At the lower ionic strengths a small correction is, however, necessary, because the concentration of bicarbonate ion arising from the dissolved carbon dioxide must be considered. The computation is made with the formula (x)(KH(:03 R) =~K i _ _ _f _ (7)

(coz + HnC03)

+

Using [KHCO, x ] instead of [HCO;] gives a small value to the term RI’/F,ln [CI]/[HCO?-] in equation 4, and this must be carried on to equation 5 . For this computation an approximate value of the ionization constant K1 is ample. For the most dilute solution the correction from this source is 0.3 mv. A plot of the PK; values for the more dilute solutions from Tables I and I1 is given in Fig. 2. Within the limit of error, about 0.2 mv., the pK; values are seen to vary linearly with the first power of the ionic strength. A larger plot would show this straight line continuing to an abscissa corresponding to = 0.23. This plot is, of course, equivalent to the formula pK:

=:

pK1

- kp

(8’1

in which pK1is the limiting value of PKi and k is a constant, with the value of 0.119, which was obtained by the method of least squares. The agreement between the observed values and those computed from equation 8 can be seen in the last two columns of Tables I and 11. Equation 8 follows from pK: = PKI - log Y H Y C I / Y H Y H C O ~

used in equation 5 if the activity coefficients follow the Debye-Hiickel formula - log

0.5048